RESEARCH

40 39 MASS EXTINCTION We measured 19 high-precision Ar/ Ar ages for the DT that complement the previous geo- chronology to create a higher-resolution tempo- ral framework for DT volcanism. We collected The eruptive tempo of Deccan samples for geochronological analysis from the region of the DT, the most rele- volcanism in relation to the vant region for understanding DT-induced climate change, as the record of the most voluminous Cretaceous-Paleogene boundary eruptive phase of the DT occurs here (Fig. 1). The total lava stratigraphy is called the Deccan Group, 1,2 1,3 4 5 which is divided into formations within the larger Courtney J. Sprain *, Paul R. Renne , Loÿc Vanderkluysen , Kanchan Pande , Stephen Self1, Tushar Mittal1 , , and Wai subgroups (in ascend- ing order) (Fig. 1). These formational and subgroup boundaries arise from geochemical and volcano- Late Cretaceous records of environmental change suggest that Deccan Traps logical properties (23). Each formation comprises (DT) volcanism contributed to the Cretaceous-Paleogene boundary (KPB) multiple eruptive units. In total, we sampled each ecosystem crisis. However, testing this hypothesis requires identification of subgroup and all but two formations within these the KPB in the DT. We constrain the location of the KPB with high-precision subgroups, including the stratigraphically high- argon-40/argon-39 data to be coincident with changes in the magmatic plumbing est and lowest dated samples from the Western system. We also found that the DT did not erupt in three discrete large pulses and Ghats. Our samples came from multiple sections. that >90% of DT volume erupted in <1 million years, with ~75% emplaced post-KPB. We focused sampling near the Lonavala–Wai Downloaded from Late Cretaceous records of climate change coincide temporally with the eruption Subgroup transition, the hypothesized location of the smallest DT phases, suggesting that either the release of climate-modifying oftheKPBbecauseofchangesinlavaflowmor- gases is not directly related to eruptive volume or DT volcanism was not the source phology, flow-field volumes, and geochemistry of Late Cretaceous climate change. ascribed to effects of the Chicxulub impact (7, 17). We collected samples to fill in the sampling gaps he mass extinction at the Cretaceous- and marine faunas (11–15). Furthermore, a tem- from Renne et al.(7)andSchoeneet al.(8), which http://science.sciencemag.org/ Paleogene boundary (KPB) fundamen- poral correlation between other flood basalt allows testing and improvement of their geo- tally reshaped Earth’s biosphere, ending eruptions and major ecological crises in Earth’s chronological models. the >150-million-year Age of the Dinosaurs history suggests the potential for the DT to have The 40Ar/39Ar method dates the eruption of T 16 and paving the way for the rise and dom- caused the mass extinction alone ( ). Additional lavas without the need for assumptions about inance of mammalian fauna. Understanding this hypotheses suggest a connection between the pre-eruptive residence time or provenance, as event is important for several reasons, including two mechanisms. Richards et al.(17)hypothesized are required for U-Pb zircon dates. We analyzed its implications for mammalian evolution and that major transitions in lava flow morphology, samples in detailed (11- to 21-step) step-heating the effects of abrupt climate change. Hypotheses flow-field volume, and feeder-dike orientation experiments with multigrain aliquots of plagi- regarding the cause of the mass extinction center observed within the DT stratigraphy were a re- oclase separates (fig. S1). To achieve high pre- around two potential triggers, invoking one or sult of a reorganization of the magmatic plumb- cision, we analyzed three to eight aliquots per

both: voluminous flood basalt volcanism (total- ing system triggered by seismic energy from the sample (fig. S1), densely bracketed by standards on February 24, 2019 ing >106 km3 of magma) from the Deccan Traps Chicxulub impact (7, 17), overall enhancing vol- during irradiation in order to precisely deter- (DT) (in modern-day ) and the large bolide canism in the DT around the time of the KPB. mine the neutron fluence (19). We report here impact recorded by the Chicxulub crater. The Central to the hypothesis that the DT played the weighted mean plateau ages for each of our impact hypothesis is supported by the Chicxulub a contributing role in the mass extinction is samples (table S1) [errors are reported at an SEM crater (which coincides in age with the main ex- the assumption that large amounts of climate- of 1-s,withX versus Y indicating analytic versus tinction event) (1) and by a global KPB impact modifying gases (CO2,CH4,andSO2) were re- systematic uncertainty, respectively, per (24)]. ejecta layer (consisting of an iridium anomaly, leased by the Deccan before the KPB. Earlier When we combined our data with previously spherules, shocked minerals, and Ni spinels) geochronological studies (10, 18, 19) suggested published high-precision dates (7), we found (2–4), in addition to evidence of abrupt extinc- that the DT erupted in three phases, with ~80% that the DT lavas erupted quasi-continuously for tion recorded by marine microfossils and terres- of the extrusive volume erupting in phase 2, a 991,000 years (see Fig. 2), from ~66.413 Ma ago trial pollen and spores [e.g., (5, 6)]. Support for short pulse starting ~400,000 years before the [the date for Jawhar Formation (Fm.) sample the Deccan hypothesis includes geochronologic KPB and ending at the boundary. Phase 2 is often KAS15-3] to ~65.422 Ma ago (the date for upper evidence that much of the DT erupted around the cited as the source of Late Cretaceous environ- Fm. sample PAN15-3). This time KPB within a time span of ~1 million years (Ma) mental change (12, 14, 20–22). With the proposed interval represents a total estimated volume of (dominantly within chron C29r) (7–10), and DT location of the KPB as suggested from the tran- ~560,000 km3 [on the basis of volume estimates volcanism roughly coincides with Late Cretaceous sitions described above, attributed to the effects presented in (17), where estimates are weighted records of climate change, in addition to records of the Chicxulub impact (7, 17), it has also been by areal extent], including ~93% of the total of ecological stress observed in some terrestrial hypothesized that >75% of the volume of the DT estimated volume of the DT. By using our com- erupted post-KPB. Current high-precision geo- posite stratigraphic section, we determined from chronology [U-Pb (8)and40Ar/39Ar (7)] cannot our data that ~85% of the Kalsubai Subgroup 1Department of Earth and Planetary Science, University of adequately test among these hypotheses because erupted in a period of 242,000 ± 101,000 years, California, Berkeley, 307 McCone Hall, Berkeley, CA 94720- of sampling gaps that fail to pinpoint the KPB that ~95% of the Lonavala Subgroup erupted 2 4767, USA. Geomagnetism Laboratory, Department of Earth, within the Deccan lava stratigraphy. To better in 46,000 ± 129,000 years, and that ~95% of the Ocean and Ecological Sciences, University of Liverpool, Liverpool L69 7ZE, UK. 3Berkeley Geochronology Center, understand the role of DT volcanism in end- Wai Subgroup erupted in a period of 690,000 ± 2455 Ridge Road, Berkeley, CA 94709, USA. 4Department of Cretaceous environmental change and mass ex- 185,000 years. Biodiversity, Earth and Environmental Science, Drexel tinction,herewereporthigh-resolution40Ar/39Ar From our age estimates for the Jawhar Fm., we University, 3245 Chestnut Street, PISB 123, Philadelphia, PA 5 plagioclase ages from the DT that locate the KPB found no evidence for older eruptions [the basis 19104, USA. Department of Earth Sciences, Indian Institute “ ” of Technology Bombay, Powai, Mumbai 400 076, India. and better refine the timing and tempo of erup- for phase 1 and the proposed Latifwadi Fm. *Corresponding author. Email: [email protected] tive fluxes. (10, 18, 19)]. Additionally, with our new data, we

Sprain et al., Science 363, 866–870 (2019) 22 February 2019 1of5 RESEARCH | REPORT Downloaded from http://science.sciencemag.org/

Fig. 1. Stratigraphy and location map. (A) Dated horizons within was adapted from (9, 10). Stratigraphy for the ABO () stratigraphic sections. Ages reported in millions of years are shown with section was adapted from (32). Stratigraphy for the MSJ () 1-s uncertainties (SEM). Ages shown in blue are new ages reported section was acquired in (7). Stratigraphic information for the KJA in this study, and ages in black were reported in (7). Sample names for ( Ghat) and VER () sections was acquired in this dated horizons are given in parentheses. Dates for MG7 and BOR14-1 study. asl, above sea level. (B) Location map of samples collected in this are the weighted mean ages combined with results presented in (7). study. (C) Sampled traverses within the Western Ghats region. Orange Stratigraphy for sections KAS (Shahapur-Igatpuri), MAT (Matheran- dots indicate samples presented in this study, and red dots indicate Neral), BOR (Khopoli-), and AMB (Mahabaleshwar-Poladpur) samples presented in (7).

on February 24, 2019 have further confirmed that the KPB is not lo- formations at the top of the Lonavala Subgroup Poladpur boundary are due to changes in the cated at the contact between the Mahabaleshwar (17), we employed the Bayesian age model “Bacon” magmatic system caused by the seismic energy and Ambenali formations but is lower in the (26). Because of uncertainties in volume estimates from the Chicxulub impact. stratigraphy and that no obvious, long hiatus and possibly diachronous formation contacts By using the above-described placement for in eruption occurred near the KPB. By con- between sections, the most secure application the KPB, we determined a mean magma extru- trast, we find an enhanced period of eruption of such models is to data from a continuous sion rate of 0.4 ± 0.1 km3/year, representing around the time of the KPB, demonstrated by stratigraphic section. Accordingly, we modeled 124,000 km3 of lava, for units erupted before rapid emplacement of the Poladpur Fm. We our data from the Ambenali Ghat section, our the KPB (comprising the Kalsubai and Lonavala thus conclude that the persistent concept of sole section containing the Bushe-Poladpur con- subgroups) and a mean extrusion rate of 0.6 ± three discrete phases of Deccan eruptions in the tact in a larger stratigraphic context. In this 0.2 km3/year, representing 435,000 km3 of lava, manner defined by Chenet et al.(10, 18)inthe section, the contact is bracketed to within 50 m for units emplaced after the KPB (comprising the Western Ghats is an artifact of limited and lower- by one date from the Poladpur Fm. and one Wai Subgroup) (Fig. 2). These results suggest precision geochronology and hence should be from the Bushe Fm. We found an interpolated that the mean extrusion rate may have increased not be used for interpreting paleoenvironmental- age of 66.03 ± 0.04 Ma (68% confidence) for after the KPB. Furthermore, using this place- proxy records (19). the Bushe-Poladpur transition (Fig. 3), which is ment of the KPB within the Deccan stratigraphy We place the KPB horizon (dated at 66.052 ± indistinguishable from the 66.052 ± 0.008 Ma indicates that >75% of the DT volume erupted 0.008/0.043 Ma via the 40Ar/39Ar technique on a 40Ar/39Ar age of the KPB (25).Theresultsof within ~650,000 years around or after the KPB. volcanic ash located 1 cm above the Ir anomaly in our age model indicate that the transition from Usinginsteadourupperboundoftheplacement eastern Montana, USA) (25) within or near the the Bushe Fm. to the Poladpur Fm. at Ambenali of the KPB near the top of the Poladpur Fm. top of the Lonavala or the basal Wai Subgroup, Ghat occurred between 60,000 years before and suggests that >50% of the erupted volume was roughly coincident with the observed transi- 20,000 years after the KPB. We cannot exclude emplaced after the KPB. Both of these results are tions that are suggested to reflect a fundamen- the possibility that the KPB occurs within the in stark contrast to previously proposed eruption tal change in the DT magmatic plumbing system. BusheorthelowerhalfofthePoladpurFm., time scales, which had ~80% of the Deccan vol- Within the uncertainty of our data, ages from but the most probable placement according to ume emplaced before the KPB (10). the Bushe Fm. through the lower Ambenali Fm. our model is ~25 m below the contact between Our results require a fundamental reassessment overlap within the uncertainty bounds of the KPB. the two. With these results, we cannot reject the oftheroleoftheDTintheKPBmassextinctionif To test the hypothesis that the KPB coincided with hypothesis that the major transitions observed the release of climate-modifying magmatic gases the transition between the Bushe and Poladpur within the Deccan stratigraphy near the Bushe- (mainly CO2 and SO2) was synchronous with the

Sprain et al., Science 363, 866–870 (2019) 22 February 2019 2of5 RESEARCH | REPORT Downloaded from http://science.sciencemag.org/

Fig. 2. Stratigraphic summary and eruptive history. Composite stratigraphic thickness (A) and cumulative volume (B) versus age for chemical formations within the DT volcanic group from the Western Ghats. All ages are plotted with 1-s uncertainties (including systematic sources).

climax of erupted lava volume, as is commonly systems have shown that a substantial quantity surface eruptive flux approaching the KPB would on February 24, 2019 assumed. Because the most voluminous Wai of magmatic volatiles can be degassed from the have led to the persistent injection of sulfur into Subgroup lava flows were emplaced around or magma reservoir at a shallow crustal depth with- the atmosphere and a pre-KPB global cooling, or after the KPB, the largest climatic impact, on the out any corresponding lava flow forming erup- conversely, silicate weathering feedbacks could basis of the above-mentioned assumption, should tions [e.g., (29) and references therein]. These havedrawndownCO2 from the atmosphere, be expected at this time as well, either just before volatiles are introduced passively into the at- also resulting in global cooling (15). We expect or in the 600,000 years after the KPB. Contrary mosphere through preexisting faults and surface the proposed pre-KPB degassing to be dominated to this expectation, proxy records show no evi- hydrological systems. The release of volatiles be- by magmatic volatiles as opposed to carbon re- dence for major climate change in the million cause of magmatic heating of crustal materials leased by the metamorphism of sedimentary years post-KPB besides a short-lived (months to such as coal has been inferred as a source for organic material, as there is no major negative millennia) pulse of cooling (2° to 4°C) immedi- CO2 related to other flood basalts [e.g., (30)], carbon isotope anomaly coinciding with the pre- ately after the KPB and a ~100,000-year record of which may be decoupled from eruptive activity. KPB warming (15), in addition to there being warming (~5°C) postextinction, which have both Whereas passive degassing of CO2 would have no evidence of organic-rich rock in the Deccan been attributed to the Chicxulub impact (Fig. 4), a global warming impact similar to that of CO2 basement. This suggests that sources of isotopi- in addition to a general similarity between the injected during an eruption, SO2 (which con- cally light carbon (e.g., biogenic methane or the duration of post-KPB volcanism and marine and verts to sulfate aerosols in the atmosphere) needs oxidation of organic matter) were not destabilized terrestrial biotic recovery (~1 Ma) (27, 28). Instead, to make it into the stratosphere in voluminous and released in substantial quantities during what is seen is that the largest climatic changes eruptions and be nearly continually replenished the DT event (15). Post-KPB, DT eruptions were occurred before the KPB, with a ~2.5° to 5°C to have a global cooling effect because of sulfur’s marked by larger eruptive events followed po- warming event around 150,000 to 450,000 years short residence time in the troposphere. tentially by longer hiatuses between eruptions before the KPB, followed by a ~5°C cooling event To reconcile pre-KPB and post-KPB climate (7). These longer recurrence intervals obviated leading up to the KPB, during the eruption of the signals with our estimated DT eruptive fluxes, we the buildup of sulfate aerosols and allowed for smallest-volume phases of the DT (11, 14, 15). It propose the following conceptual model (Fig. 4). CO2 drawdown through silicate weathering and follows either that the DT was not the cause of The initiation of the Deccan mantle plume and organic carbon burial, leading to lower levels of Late Cretaceous climate change and did not play the initial emplacement of magma in the litho- accumulated atmospheric CO2 partial pressure. aroleinthemassextinctionorthatthereleaseof sphere and cold, upper crust led to degassing of In addition to longer recurrence intervals, early- climate-modifying gases is not directly related to a large amount of magmatic volatiles (especially stage passive degassing may have reduced the erupted lava volume, as previously assumed. CO2) before the surface eruptions resulting in volatile content in magmas erupted after the Over the past few decades, increased monitor- global warming (31). Subsequently, if recurrence KPB, which may have been pushed to eruption ing and measurements at present-day volcanic times were short enough, the increase in the DT by processes related to the Chicxulub impact.

Sprain et al., Science 363, 866–870 (2019) 22 February 2019 3of5 RESEARCH | REPORT

eruptions and passively, and climatic effects of Fig. 3. Bayesian age model for large igneous provinces need to be obtained. the Ambenali Ghat section. Results indicate an interpolated age of 66.03 ± 0.04 Ma (68% REFERENCES AND NOTES confidence) for the Bushe- 1. P. R. Renne et al., Science 339, 684–687 (2013). Poladpur transition, which is 2. L. W. Alvarez, W. Alvarez, F. Asasro, H. V. Michel, Science 208, 1095–1108 (1980). indistinguishable from the 3. A. Montanari et al., Geology 11, 668–671 (1983). 66.052 ± 0.008 Ma 40Ar/39Ar 4. B. F. Bohor, Tectonophysics 171, 359–372 (1990). age of the KPB (25). The Bayesian 5. J. Smit, Geol. Mijnb. 69, 187–204 (1990). – age model Bacon (26) was used 6. A. R. Sweet, D. R. Braman, Can. J. Earth Sci. 38,249 269 (2001). 19 ( ). Age uncertainties (red 7. P. R. Renne et al., Science 350,76–78 (2015). ellipses) are shown at 2-s and 8. B. Schoene et al., Science 347, 182–184 (2015). exclude external sources. masl, 9. A. L. Chenet, F. Fluteau, V. Courtillot, M. Gérard, K. V. Subbarao, meters above sea level. J. Geophys. Res. Solid Earth 113,B04101(2008). 10. A. L. Chenet et al., J. Geophys. Res. Solid Earth 114, B06103 (2009). 11. T. S. Tobin, G. P. Wilson, J. M. Eiler, J. H. Hartman, Geology 42, 351–354 (2014). 12. G. P. Wilson, D. G. DeMar Jr., G. Carter, “Extinction and survival of salamander and salamander-like amphibians across the Cretaceous-Paleogene boundary in northeastern Montana, USA,” in Through the End of the Cretaceous in the Type Locality Downloaded from of the Hell Creek Formation in Montana and Adjacent Areas, G. P. Wilson, W. A. Clemens, J. R. Horner, J. H. Hartman, Eds. (GSA Special Papers, vol. 503, Geological Society of America, 2014), pp. 271–297. 13. M. Kucera, B. A. Malmgren, Paleobiology 24,49–63 (1998). 14. T. S. Tobin et al., Palaeogeogr. Palaeoclimatol. Palaeoecol. 350–352, 180–188 (2012). 15. J. S. K. Barnet et al., Geology 46,147–150 (2017). 16. V. E. Courtillot, P. R. Renne, C. R. Geosci. 335,113–140 http://science.sciencemag.org/ (2003). 17. M. A. Richards et al., Geol.Soc.Am.Bull.127,1507–1520 (2015). 18. A. L. Chenet, X. Quidelleur, F. Fluteau, V. Courtillot, S. Bajpai, Earth Planet. Sci. Lett. 263,1–15 (2007). 19. See supplementary materials. 20. G. Keller, T. Adatte, S. Gardin, A. Bartolini, S. Bajpai, Earth Planet. Sci. Lett. 268, 293–311 (2008). 21. G. P. Wilson, “Mammalian extinction, survival, and recovery dynamics across the Cretaceous-Paleogene boundary in northeastern Montana, USA,” in Through the End of the Cretaceous in the Type Locality of the Hell Creek Formation in

Montana and Adjacent Areas, G. P. Wilson, W. A. Clemens, on February 24, 2019 J. R. Horner, J. H. Hartman, Eds. (GSA Special Papers, vol. 503, Geological Society of America, 2014), pp. 365–392. 22. S. V. Petersen, A. Dutton, K. C. Lohmann, Nat. Commun. 7, 12079 (2016). 23. J. E. Beane, C. A. Turner, P. R. Hooper, K. V. Subbarao, J. N. Walsh, Bull. Volcanol. 48,61–83 (1986). 24. P. R. Renne, G. Balco, K. R. Ludwig, R. Mundil, K. Min, Geochim. Cosmochim. Acta 75, 5097–5100 (2011). 25. C. J. Sprain, P. R. Renne, W. A. Clemens, G. P. Wilson, Geol. Soc. Am. Bull. 130, 1615–1644 (2018). 26. M. Blaauw, J. A. Christeny, Bayesian Anal. 6, 457–474 (2011). 27. J. Vellekoop et al., Proc. Natl. Acad. Sci. U.S.A. 111, 7537–7541 (2014). 28. K. G. MacLeod, P. C. Quinton, J. Sepúlveda, M. H. Negra, Science 360, 1467–1469 (2018). 29. M. Edmonds, P. J. Wallace, Elements 13,29–34 (2017). 30. B. A. Black, L. T. Elkins-Tanton, M. C. Rowe, I. U. Peate, Earth Planet. Sci. Lett. 317–318, 363–373 (2012). 31. S. V. Sobolev et al., Nature 477, 312–316 (2011). Fig. 4. Eruptive flux and climatic changes. Correlation of Deccan eruptive fluxes to benthic d18O 32. C. W. Devey, P. C. Lightfoot, Bull. Volcanol. 48,195–207 15 (1986). data from Ocean Drilling Program site 1262 (blue line) [published in ( )]. Colored blocks represent 33. A. E. Jay, C. M. Niocaill, M. Widdowson, S. Self, W. Turner, eruptive fluxes, where color indicates the formation per Fig. 2; horizontal length indicates the J. Geol. Soc. London 166,13–24 (2009). approximate duration; and height is scaled by eruptive volume as calculated in (17). Red lines mark 34. J. Bowles, Proc. Ocean Drill. Program Sci. Results 208,1–24 the known locations of red boles taken from (9, 10, 33). Magnetozones for the oxygen isotope (2006). data (Deccan) are from (9, 10, 33, 34). Ages shown for the KPB, C29r/C29n, and C30n/C29r reversals ACKNOWLEDGMENTS 40 39 25 T are Ar/ Ar ages from ( ). , temperature. We thank T. Becker and A. Jaouni for laboratory assistance; H. Sheth, R. Duraiswami, V. Kale, I. Fendley, and M. Richards for field assistance; and W. Alvarez, M. Manga, B. Black, and Although the above-described model helps to better understanding of its volatile release is M. Richards for discussion. Funding: This work was funded by reconcile our calculated DT eruptive fluxes with required. In particular, better estimates of the the Ann and Gordon Getty Foundation; the Esper S. Larsen Fund observed climate changes around the KPB, to amount and specific species of gas released and of the University of California, Berkeley; the Heising-Simons Foundation; NSF grants EAR-1615021, EAR-1615203, and EAR-1615003; truly understand the specific role Deccan vol- an improved understanding of the relationships and the Berkeley Geochronology Center. C.J.S. was supported by canism played in the KPB mass extinction, a among lava extrusion, volatile release both during an NSF graduate research fellowship. Author contributions:

Sprain et al., Science 363, 866–870 (2019) 22 February 2019 4of5 RESEARCH | REPORT

C.J.S. collected samples, processed and analyzed samples, and materials availability: Alldataareavailableinthe Supplementary Text interpreted results, and wrote and edited the manuscript; P.R.R. supplementary materials. Fig. S1 collected samples, processed and analyzed samples, and Tables S1 and S2 participated in interpretation and the writing and editing of References (35–44) the manuscript; L.V., S.S., K.P., and T.M. collected samples SUPPLEMENTARY MATERIALS and participated in interpretation and the writing and editing of www.sciencemag.org/content/363/6429/866/suppl/DC1 23 August 2018; accepted 8 January 2019 the manuscript. Competing interests: None declared. Data Materials and Methods 10.1126/science.aav1446 Downloaded from http://science.sciencemag.org/

on February 24, 2019

Sprain et al., Science 363, 866–870 (2019) 22 February 2019 5of5